A breakthrough in semiconductor technology (computing) may be on the horizon with the creation of atomically tunable “memristors”—advanced memory resistors designed to mimic the neural networks of the human brain.
Supported by the National Science Foundation’s Future of Semiconductors program (FuSe2), this project seeks to develop devices that enable neuromorphic computing—a cutting-edge approach to processing that mirrors the brain’s ability to learn, adapt, and operate with remarkable energy efficiency.
At the heart of this innovation lies the development of ultrathin memory devices with atomic-scale precision. These devices, which function as artificial synapses and neurons, hold the potential to transform artificial intelligence by delivering unparalleled computing power and efficiency. The implications for AI applications are immense, and the initiative is also geared toward training a new generation of semiconductor technology experts.
Addressing Neuromorphic Computing Challenges
The initiative tackles a key challenge in modern computing: achieving the accuracy and scalability required to bring brain-inspired AI systems to life. To replicate the human brain’s ability to perform parallel data processing efficiently, memristors are essential. These devices can simultaneously store and process information, making them ideal for neuromorphic circuits. By leveraging this parallelism, memristors could overcome the limitations of traditional computing architectures, paving the way for energy-efficient, high-speed neural networks.
Pioneering Atomic-Scale Memory Technology
This collaborative research effort, led by Judy Wu, a distinguished professor of Physics and Astronomy at the University of Kansas (KU), is a partnership between KU and the University of Houston. Backed by a $1.8 million FuSe2 grant, the team is exploring groundbreaking advancements in memory device design.
Wu’s team has developed a process to achieve memory devices with sub-2-nanometer thickness, with film layers as thin as 0.1 nanometers—nearly 10 times thinner than conventional nanometer-scale devices. These ultrathin designs represent a significant step forward for future semiconductor electronics, enabling devices that are both exceptionally thin and precisely functional. The researchers are also employing a co-design approach, integrating material engineering, device fabrication, and performance testing.
Building the Future Workforce
Beyond its technical achievements, the project emphasizes workforce development to address the increasing demand for skilled professionals in the semiconductor industry. To this end, the team has included an educational outreach program led by experts from both universities.
“Our primary goal is to develop atomically tunable memristors that function as neurons and synapses in neuromorphic circuits, enabling neuromorphic computing,” Wu explained. “We aim to replicate how the brain thinks, computes, makes decisions, and recognizes patterns—all with incredible speed and energy efficiency.”
